Professional Documents
Culture Documents
Contents
1. Introduction
2. Basic notions
3. Preadditive and additive categories
4. Karoubian categories
5. Abelian categories
6. Extensions
7. Additive functors
8. Localization
9. Serre subcategories
10. K-groups
11. Cohomological delta-functors
12. Complexes
13. Truncation of complexes
14. Homotopy and the shift functor
15. Graded objects
16. Filtrations
17. Spectral sequences
18. Spectral sequences: exact couples
19. Spectral sequences: differential objects
20. Spectral sequences: filtered differential objects
21. Spectral sequences: filtered complexes
22. Spectral sequences: double complexes
23. Injectives
24. Projectives
25. Injectives and adjoint functors
26. Essentially constant systems
27. Inverse systems
28. Exactness of products
29. Other chapters
References
1
2
2
4
5
12
13
16
18
20
22
24
28
30
33
34
40
41
43
45
47
50
55
55
56
58
59
63
63
64
1. Introduction
Basic homological algebra will be explained in this document. We add as needed
in the other parts, since there is clearly an infinite amount of this stuff around. A
reference is [ML63].
This is a chapter of the Stacks Project, version 4287507, compiled on Oct 20, 2014.
1
HOMOLOGICAL ALGEBRA
2. Basic notions
The following notions are considered basic and will not be defined, and or proved.
This does not mean they are all necessarily easy or well known.
(1) Nothing yet.
3. Preadditive and additive categories
Here is the definition of a preadditive category.
Definition 3.1. A category A is called preadditive if each morphism set MorA (x, y)
is endowed with the structure of an abelian group such that the compositions
Mor(x, y) Mor(y, z) Mor(x, z)
are bilinear. A functor F : A B of preadditive categories is called additive if and
only if F : Mor(x, y) Mor(F (x), F (y)) is a homomorphism of abelian groups for
all x, y Ob(A).
In particular for every x, y there exists at least one morphism x y, namely the
zero map.
Lemma 3.2. Let A be a preadditive category. Let x be an object of A. The
following are equivalent
(1) x is an initial object,
(2) x is a final object, and
(3) idx = 0 in MorA (x, x).
Furthermore, if such an object 0 exists, then a morphism : x y factors through
0 if and only if = 0.
Proof. Omitted.
/x
?
?z
1
/ y
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
W 7 Ker(MorD (W, X)
MorD (W, X))
if representable if and only if e has a kernel. Note that for any abelian group A
and idempotent endomorphism e : A A we have
Y
Y
Ker(e : A A) = Ker( :
A
A)
nN
nN
where
(a1 , a2 , a3 , . . .) = (ea1 + (1 e)a2 , ea2 + (1 e)a3 , . . .)
Moreover, has the right inverse
(a1 , a2 , a3 , . . .) = (a1 , (1 e)a1 + ea2 , (1 e)a2 + ea3 , . . .).
Hence (1) holds. The proof of (2) is dual (using the dual definition of a Karoubian
category, namely condition (2) of Lemma 4.2).
5. Abelian categories
An abelian category is a category satisfying just enough axioms so the snake lemma
holds. An axiom (that is sometimes forgotten) is that the canonical map Coim(f )
Im(f ) of Lemma 3.11 is always an isomorphism. Example 3.12 shows that it is
necessary.
Definition 5.1. A category A is abelian if it is additive, if all kernels and cokernels
exist, and if the natural map Coim(f ) Im(f ) is an isomorphism for all morphisms
f of A.
Lemma 5.2. Let A be a preadditive category. The additions on sets of morphisms
make Aopp into a preadditive category. Furthermore, A is additive if and only if
Aopp is additive, and A is abelian if and only if Aopp is abelian.
Proof. Omitted.
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
/y
x
/ z
be a commutative diagram.
(1) The diagram is cartesian if and only if
(g,f )
(k,h)
0 w x y z
is exact.
(2) The diagram is cocartesian if and only if
(g,f )
(k,h)
w x y z 0
is exact.
Proof. Let u = (g, f ) : w x y and v = (k, h) : x y z. Let p : x y x
and q : x y y be the canonical projections. Let i : Ker(v) x y be the
canonical injection. By Example 5.6, the diagram is cartesian if and only if there
exists an isomorphism r : Ker(v) w with f r = q i and g r = p i. The
u
v
sequence 0 w x y z is exact if and only if there exists an isomorphism
r : Ker(v) w with u r = i. But given r : Ker(v) w, we have f r = q i and
g r = p i if and only if q u r = f r = q i and p u r = g r = p i, hence
if and only if u r = i. This proves (1), and then (2) follows by duality.
Lemma 5.12. Let A be an abelian category. Let
w
/y
x
/ z
be a commutative diagram.
(1) If the diagram is cartesian, then the morphism Ker(f ) Ker(k) induced
by g is an isomorphism.
(2) If the diagram is cocartesian, then the morphism Coker(f ) Coker(k)
induced by h is an isomorphism.
Proof. Suppose the diagram is cartesian. Let e : Ker(f ) Ker(k) be induced by
g. Let i : Ker(f ) w and j : Ker(k) x be the canonical injections. There exists
t : Ker(k) w with f t = 0 and g t = j. Hence, there exists u : Ker(k) Ker(f )
with iu = t. It follows giue = gte = je = gi and f iue = 0 = f i, hence
i u e = i. Since i is a monomorphism this implies u e = idKer(f ) . Furthermore,
we have j e u = g i u = g t = j. Since j is a monomorphism this implies
e u = idKer(k) . This proves (1). Now, (2) follows by duality.
HOMOLOGICAL ALGEBRA
/y
x
/ z
be a commutative diagram.
(1) If the diagram is cartesian and k is an epimorphism, then the diagram is
cocartesian and f is an epimorphism.
(2) If the diagram is cocartesian and g is a monomorphism, then the diagram
is cartesian and h is a monomorphism.
Proof. Suppose the diagram is cartesian and k is an epimorphism. Let u = (g, f ) :
w x y and let v = (k, h) : x y z. As k is an epimorphism, v is an
u
v
epimorphism, too. Therefore and by Lemma 5.11, the sequence 0 w x y
z 0 is exact. Thus, the diagram is cocartesian by Lemma 5.11. Finally, f is an
epimorphism by Lemma 5.12 and Lemma 5.4. This proves (1), and (2) follows by
duality.
Lemma 5.14. Let A be an abelian category.
(1) If x y is surjective, then for every z y the projection x y z z is
surjective.
(2) If x y is injective, then for every x z the morphism z z qx y is
injective.
Proof. Immediately from Lemma 5.4 and Lemma 5.13.
/y
u
/ v
/z
/ w
HOMOLOGICAL ALGEBRA
be a commutative diagram.
(1) If the first row is exact and k is a monomorphism, then the induced sequence
Ker() Ker() Ker() is exact.
(2) If the second row is exact and g is an epimorphism, then the induced sequence Coker() Coker() Coker() is exact.
Proof. Suppose the first row is exact and k is a monomorphism. Let a : Ker()
Ker() and b : Ker() Ker() be the induced morphisms. Let h : Ker()
x, i : Ker() y and j : Ker() z be the canonical injections. As j is a
monomorphism we have b a = 0. Let c : s Ker() with b c = 0. Then,
gic = j bc = 0. By Lemma 5.15 there are an object t, an epimorphism d : t s
and a morphism e : t x with icd = f e. Then, ke = f e = icd = 0.
As k is a monomorphism we get e = 0. So, there exists m : t Ker() with
hm = e. It follows iam = f hm = f e = icd. As i is a monomorphism we
get am = cd. Thus, Lemma 5.15 implies (1), and then (2) follows by duality.
Lemma 5.17. Let A be an abelian category. Let
/y
/ u
/ v
/0
/z
/ w
v
/ 0
y z Ker()
/ Ker()
Coker() qu v o
Coker()
commutes, where and 0 are the canonical projections and and 0 are
the canonical coprojections.
(2) The induced sequence
f0
g0
k0
l0
10
HOMOLOGICAL ALGEBRA
/p
/ Ker()
/ z
/0
/ v
/ w
/ Coker()
/ u
/ y
/0
/ q
Lemma 5.15 implies that Ker() Ker() Coker() is exact. Then, the claim
follows by Lemma 5.16 and duality.
Lemma 5.18. Let A be an abelian category. Let
/y
x
x0
/ u0
/ u
/ y0
0
/ v0
/ v
/z
/ z0
~
/ w0
/0
/ w
/0
HOMOLOGICAL ALGEBRA
11
/ Ker()
/ Ker()
/ Coker()
/ Coker()
/ Coker()
Ker(0 )
/ Ker( 0 )
/ Ker( 0 )
/ Coker(0 )
/ Coker( 0 )
/ Coker( 0 )
commutes.
Proof. Omitted.
/ x0
w0
/z
/ y0
/ z0
/ y0
/ z0
/ x0
/ Ker(y 0 z 0 )
to conclude that x x0 is surjective. This proves (1). The proof of (2) is dual to
this.
Lemma 5.20. Let A be an abelian category. Let
/x
/y
/w
v
v0
/ w0
/ x0
/z
/ y0
/ z0
12
HOMOLOGICAL ALGEBRA
6. Extensions
Definition 6.1. Let A be an abelian category. Let A, C Ob(A). An extension
E of B by A is a short exact sequence
0 A E B 0.
By abuse of language we often omit mention of the morphisms A E and E B,
although they are definitively part of the structure of an extension.
Definition 6.2. Let A be an abelian category. Let A, B Ob(A). The set of
isomorphism classes of extensions of B by A is denoted
ExtA (B, A).
This is called the Ext-group.
This definition works, because by our conventions A is a set, and hence ExtA (B, A)
is a set. In any of the cases of big abelian categories listed in Categories, Remark
2.2. one can check by hand that ExtA (B, A) is a set as well. Also, we will see
later that this is always the case when A has either enough projectives or enough
injectives. Insert future reference here.
Actually we can turn ExtA (, ) into a functor
Aopp A Sets,
as follows:
(1) Given a morphism B 0 B and an extension E of B by A we define
E 0 = E B B 0 so that we have the following commutative diagram of short
exact sequences
0
/A
/ E0
/ B0
/0
/A
/E
/B
/0
HOMOLOGICAL ALGEBRA
13
/ AA
/ E1 E2
/ BB
/0
/ E0
O
/ BB
O
/0
/A
O
/A
/ E1 + E2
/B
/0
/ HomA (M3 , N )
ExtA (M3 , N )
/ HomA (M2 , N )
/ HomA (M1 , N )
/ ExtA (M2 , N )
/ ExtA (M1 , N )
/ HomA (N, M1 )
ExtA (N, M1 )
/ HomA (N, M2 )
/ HomA (N, M3 )
/ ExtA (N, M2 )
/ ExtA (N, M3 )
14
HOMOLOGICAL ALGEBRA
(3) If F is additive then it is right exact if and only if for every short exact
sequence 0 A B C 0 the sequence F (A) F (B) F (C) 0
is exact.
(4) If F is additive then it is exact if and only if for every short exact sequence
0 A B C 0 the sequence 0 F (A) F (B) F (C) 0 is
exact.
Proof. Let us first note that if F commutes with the empty limit or the empty
colimit, then F (0) = 0. In particular F applied to the zero morphism is zero. We
will use this below without mention.
Suppose that F is left exact, i.e., commutes with finite limits. Then F (A A) =
F (A) F (A) with projections F (p) and F (q). Hence F (A A) = F (A) F (A)
with all four morphisms F (i), F (j), F (p), F (q) equal to their counterparts in B as
they satisfy the same relations, see Remark 3.6. Then f = F (p + q) is a morphism
f : F (A) F (A) F (A) such that f F (i) = F (p i + q i) = F (idA ) = idF (A) .
And similarly f F (j) = idA . We conclude that F (p + q) = F (p) + F (q). For any
pair of morphisms a, b : B A the map g = F (i a + j b) : F (B) F (A) F (A)
is a morphism such that F (p) g = F (p (i a + j b)) = F (a) and similarly
F (q) g = F (b). Hence g = F (i) F (a) + F (j) F (b). The sum of a and b is the
composition
B
ia+jb
/ AA
p+q
/ A.
Applying F we get
F (B)
/ F (A) F (A)
F (p)+F (q)
/ A.
where we used the expressions for f and g obtained above. Hence F is additive.1
Denote f : B C a map from B to C. Exactness of 0 A B C just means
that A = Ker(f ). Clearly the kernel of f is the equalizer of the two maps f and
0 from B to C. Hence if F commutes with limits, then F (Ker(f )) = Ker(F (f ))
which exactly means that 0 F (A) F (B) F (C) is exact.
Conversely, suppose that F is additive and transforms any short exact sequence
0 A B C into an exact sequence 0 F (A) F (B) F (C). Because it
is additive it commutes with direct sums and hence finite products in A. To show
it commutes with finite limits it therefore suffices to show that it commutes with
equalizers. But equalizers in an abelian category are the same as the kernel of the
difference map, hence it suffices to show that F commutes with taking kernels. Let
f : A B be a morphism. Factor f as A I B with f 0 : A I surjective
and i : I B injective. (This is possible by the definition of an abelian category.)
Then it is clear that Ker(f ) = Ker(f 0 ). Also 0 Ker(f 0 ) A I 0 and
0 I B B/I 0 are short exact. By the condition imposed on F we see
that 0 F (Ker(f 0 )) F (A) F (I) and 0 F (I) F (B) F (B/I) are exact.
Hence it is also the case that F (Ker(f 0 )) is the kernel of the map F (A) F (B),
and we win.
The proof of (3) is similar to the proof of (2). Statement (4) is a combination of
(2) and (3).
1Im sure there is an infinitely slicker proof of this.
HOMOLOGICAL ALGEBRA
15
The following lemma is used in the proof that the category of abelian sheaves on a
site is abelian, where the functor b is sheafification.
Lemma 7.3. Let a : A B and b : B A be functors. Assume that
(1) A, B are additive categories, a, b are additive functors, and a is right adjoint
to b,
(2) B is abelian and b is left exact, and
(3) ba
= idA .
Then A is abelian.
Proof. As B is abelian we see that all finite limits and colimits exist in B by Lemma
5.5. Since b is a left adjoint we see that b is also right exact and hence exact, see
Categories, Lemma 24.5. Let : B1 B2 be a morphism of B. In particular, if
K = Ker(B1 B2 ), then K is the equalizer of 0 and and hence bK is the equalizer
of 0 and b, hence bK is the kernel of b. Similarly, if Q = Coker(B1 B2 ), then
Q is the coequalizer of 0 and and hence bQ is the coequalizer of 0 and b, hence
bQ is the cokernel of b. Thus we see that every morphism of the form b in A has
a kernel and a cokernel. However, since ba
= id we see that every morphism of A
is of this form, and we conclude that kernels and cokernels exist in A. In fact, the
argument shows that if : A1 A2 is a morphism then
Ker() = bKer(a),
and
Coker() = bCoker(a).
Now we still have to show that Coim() = Im(). We do this as follows. First
note that since A has kernels and cokernels it has all finite limits and colimits (see
proof of Lemma 5.5). Hence we see by Categories, Lemma 24.5 that a is left exact
and hence transforms kernels (=equalizers) into kernels.
Coim() = Coker(Ker() A1 )
by definition
= bCoker(a(Ker() A1 ))
by formula above
= bCoker(Ker(a) aA1 ))
a preserves kernels
= bCoim(a)
= bIm(a)
= bKer(aA2 Coker(a))
= Ker(baA2 bCoker(a))
= Ker(A2 bCoker(a))
= Ker(A2 Coker())
= Im()
Thus the lemma holds.
by definition
B is abelian
by definition
b preserves kernels
ba = idA
by formula above
by definition
16
HOMOLOGICAL ALGEBRA
8. Localization
In this section we note how Gabriel-Zisman localization interacts with the additive
structure on a category.
Lemma 8.1. Let C be a preadditive category. Let S be a left or right multiplicative system. There exists a canonical preadditive structure on S 1 C such that the
localization functor Q : C S 1 C is additive.
Proof. We will prove this in the case S is a left multiplicative system. The case
where S is a right multiplicative system is dual. Suppose that X, Y are objects of
C and that , : X Y are morphisms in S 1 C. According to Categories, Lemma
25.3 we may represent these by pairs s1 f, s1 g with common denominator s. In
this case we define + to be the equivalence class of s1 (f + g). In the rest of
the proof we show that this is well defined and that composition is bilinear. Once
this is done it is clear that Q is an additive functor.
Let us show construction above is well defined. An abstract way of saying this is
that filtered colimits of abelian groups agree with filtered colimits of sets and to use
Categories, Equation (25.5.1). We can work this out in a bit more detail as follows.
Say s : Y Y1 and f, g : X Y1 . Suppose we have a second representation of
, as (s0 )1 f 0 , (s0 )1 g 0 with s0 : Y Y2 and f 0 , g 0 : X Y2 . By Categories,
Remark 25.5 we can find a morphism s3 : Y Y3 and morphisms a1 : Y1 Y3 ,
a2 : Y2 Y3 such that a1 s = s3 = a2 s0 and also a1 f = a2 f 0 and a1 g = a2 g 0 .
Hence we see that s1 (f + g) is equivalent to
1
s1
3 (a1 (f + g)) = s3 (a1 f + a1 g)
0
0
= s1
3 (a2 f + a2 g )
0
0
= s1
3 (a2 (f + g ))
/ Z0
/ Z 00
t0
f,g
/ Y0
Then = (t0 t)1 (a f ) and = (t0 t)1 (a g). Hence we see that
( + ) is represented by (t0 t)1 (a (f + g)) = (t0 t)1 (a f + a g) which
represents + .
Finally, assume that : W X is another morphism of S 1 C. Say = r1 i for
some i : W X 0 and r : X X 0 in S. We claim that we can find a morphism
HOMOLOGICAL ALGEBRA
17
X
s
f,g,f +g
/ Y0
s0
18
HOMOLOGICAL ALGEBRA
0 Ob(C),
C is a strictly full subcategory of A,
any subobject or quotient of an object of C is an object of C,
if A Ob(A) is an extension of objects of C then also A Ob(C).
HOMOLOGICAL ALGEBRA
19
C
/B
/ C qA B
20
HOMOLOGICAL ALGEBRA
AOb(A)
...
/M
/M
/M
/M
/ ...
HOMOLOGICAL ALGEBRA
21
X
iI +
X
iI
iI
iI
iI
iI +
[Bi ].
Since the right and left hand side should contain the same objects of A counted
with multiplicity, this means there should be a bijection between the terms which
occur above. Set
a
a
T + = {p}
{a, c} I +
{b} I
and
T = {q}
{a, c} I
{b} I + .
22
Set T = T +
HOMOLOGICAL ALGEBRA
T = {p, q}
`
{a, b, c} I.
Q
O(t) = Ai
Bi
Ci
For t T define
if
t=p
if
t=q
if t = (a, i)
if t = (b, i)
if t = (c, i)
Hence we can view : T + T as a bijection such that O(t) = O( (t)) for all
+
+
t T + . Let t
be the unique element such that (t+
0 = (p) and let t0 T
0 ) = q.
Consider the object
M
O(t)
M+ =
+
tT
tT
HOMOLOGICAL ALGEBRA
23
/ F 0 (A)
/ F 0 (B)
u
ABC
/ F 1 (B)
F (A)
F (A)
/ F 0 (C)
/ F 1 (C)
ABC
/ F 2 (B)
/ ...
is exact, and
(2) for every morphism (A B C) (A0 B 0 C 0 ) of short exact
sequences of A the diagrams
F n (C)
F n (C 0 )
ABC
A0 B 0 C 0
/ F n+1 (A)
/ F n+1 (A0 )
are commutative.
Note that this in particular implies that F 0 is left exact.
Definition 11.2. Let A, B be abelian categories. Let (F n , F ) and (Gn , G ) be
-functors from A to B. A morphism of -functors from F to G is a collection of
transformation of functors tn : F n Gn , n 0 such that for every short exact
sequence 0 A B C 0 of A the diagrams
F n (C)
F,ABC
tn
Gn (C)
G,ABC
/ F n+1 (A)
tn+1
/ Gn+1 (A)
are commutative.
Definition 11.3. Let A, B be abelian categories. Let F = (F n , F ) be a -functor
from A to B. We say F is a universal -functor if an only if for every -functor
G = (Gn , G ) and any morphism of functors t : F 0 G0 there exists a unique
morphism of -functors {tn }n0 : F G such that t = t0 .
Lemma 11.4. Let A, B be abelian categories. Let F = (F n , F ) be a -functor from
A to B. Suppose that for every n > 0 and any A Ob(A) there exists an injective
morphism u : A B (depending on A and n) such that F n (u) : F n (A) F n (B)
is zero. Then F is a universal -functor.
Proof. Let G = (Gn , G ) be a -functor from A to B and let t : F 0 G0 be
a morphism of functors. We have to show there exists a unique morphism of functors {tn }n0 : F G such that t = t0 . We construct tn by induction on n.
For n = 0 we set t0 = t. Suppose we have already constructed a unique sequence of
transformation of functors ti for i n compatible with the maps in degrees n.
Let A Ob(A). By assumption we may choose a embedding u : A B such
that F n+1 (u) = 0. Let C = B/u(A). The long exact cohomology sequence for
24
HOMOLOGICAL ALGEBRA
tn
tn+1
A
F n+1 (A)
G,ABC
/ Gn+1 (A)
12. Complexes
Of course the notions of a chain complex and a cochain complex are dual and you
only have to read one of the two parts of this section. So pick the one you like.
(Actually, this doesnt quite work right since the conventions on numbering things
are not adapted to an easy transition between chain and cochain complexes.)
A chain complex A in an additive category A is a complex
dn+1
n
An1 . . .
. . . An+1 An
of A. In other words, we are given an object Ai of A for all i Z and for all i Z
a morphism di : Ai Ai1 such that di1 di = 0 for all i. A morphism of chain
complexes f : A B is given by a family of morphisms fi : Ai Bi such that
all the diagrams
/ Ai1
Ai
di
fi
Bi
di
fi1
/ Bi1
commute. The category of chain complexes of A is denoted Ch(A). The full subcategory consisting of objects of the form
. . . A2 A1 A0 0 0 . . .
is denoted Ch0 (A). In other words, a chain complex A belongs to Ch0 (A) if
and only if Ai = 0 for all i < 0. A homotopy h between a pair of morphisms of
chain complexes f, g : A B is is a collection of morphisms hi : Ai Bi+1 such
that we have
fi gi = di+1 hi + hi1 di
for all i. Clearly, the notions of chain complex, morphism of chain complexes,
and homotopies between morphisms of chain complexes makes sense even in a
preadditive category.
HOMOLOGICAL ALGEBRA
25
In particular this means that it makes sense to define the category of chain complexes with maps up to homotopy. Well return to this later.
Definition 12.2. Let A be an additive category. We say a morphism a : A B
is a homotopy equivalence if there exists a morphism b : B A such that there
exists a homotopy between a b and idA and there exists a homotopy between b a
and idB . If there exists such a morphism between A and B , then we say that A
and B are homotopy equivalent.
In other words, two complexes are homotopy equivalent if they become isomorphic
in the category of complexes up to homotopy.
Lemma 12.3. Let A be an abelian category.
(1) The category of chain complexes in A is abelian.
(2) A morphism of complexes f : A B is injective if and only if each
fn : An Bn is injective.
(3) A morphism of complexes f : A B is surjective if and only if each
fn : An Bn is surjective.
(4) A sequence of chain complexes
f
A
B
C
is exact at B if and only if each sequence
fi
gi
Ai Bi Ci
is exact at Bi .
Proof. Omitted.
For any i Z the ith homology group of a chain complex A in an abelian category
is defined by the following formula
Hi (A ) = Ker(di )/Im(di+1 ).
If f : A B is a morphism of chain complexes of A then we get an induced
morphism Hi (f ) : Hi (A ) Hi (B ) because clearly fi (Ker(di : Ai Ai1 ))
Ker(di : Bi Bi1 ), and similarly for Im(di+1 ). Thus we obtain a functor
Hi : Ch(A) A.
Definition 12.4. Let A be an abelian category.
(1) A morphism of chain complexes f : A B is called a quasi-isomorphism
if the induced maps Hi (f ) : Hi (A ) Hi (B ) is an isomorphism for all
i Z.
(2) A chain complex A is called acyclic if all of its homology objects Hi (A )
are zero.
Lemma 12.5. Let A be an abelian category.
26
HOMOLOGICAL ALGEBRA
Hi (A )
Hi1 (A )
...
...
/ Hi (B )
/ Hi (C )
/ Hi1 (B )
/ Hi1 (C )
...
...
... s
Proof. Omitted. The maps come from the Snake Lemma 5.17 applied to the
diagrams
/ Bi /Im(dB,i+1 )
Ai /Im(dA,i+1 )
dA,i
/ Ker(dA,i1 )
/ Ci /Im(dC,i+1 )
dB,i
/ Ker(dB,i1 )
/0
dC,i
/ Ker(dC,i1 )
dn
. . . An1 An An+1 . . .
of A. In other words, we are given an object Ai of A for all i Z and for all i Z
a morphism di : Ai Ai+1 such that di+1 di = 0 for all i. A morphism of cochain
complexes f : A B is given by a family of morphisms f i : Ai B i such that
all the diagrams
/ Ai+1
Ai
di
f i+1
di /
Bi
B i+1
commute. The category of cochain complexes of A is denoted CoCh(A). The full
subcategory consisting of objects of the form
. . . 0 0 A0 A1 A2 . . .
is denoted CoCh0 (A). In other words, a cochain complex A belongs to the
subcategory CoCh0 (A) if and only if Ai = 0 for all i < 0. A homotopy h between
HOMOLOGICAL ALGEBRA
27
In particular this means that it makes sense to define the category of cochain
complexes with maps up to homotopy. Well return to this later.
Definition 12.8. Let A be an additive category. We say a morphism a : A B
is a homotopy equivalence if there exists a morphism b : B A such that there
exists a homotopy between a b and idA and there exists a homotopy between b a
and idB . If there exists such a morphism between A and B , then we say that A
and B are homotopy equivalent.
In other words, two complexes are homotopy equivalent if they become isomorphic
in the category of complexes up to homotopy.
Lemma 12.9. Let A be an abelian category.
(1) The category of cochain complexes in A is abelian.
(2) A morphism of cochain complexes f : A B is injective if and only if
each f n : An B n is injective.
(3) A morphism of cochain complexes f : A B is surjective if and only if
each f n : An B n is surjective.
(4) A sequence of cochain complexes
f
A
B
C
is exact at B if and only if each sequence
fi
gi
Ai B i C i
is exact at B i .
Proof. Omitted.
For any i Z the ith cohomology group of a cochain complex A is defined by the
following formula
H i (A ) = Ker(di )/Im(di1 ).
If f : A B is a morphism of cochain complexes of A then we get an induced
morphism H i (f ) : H i (A ) H i (B ) because clearly f i (Ker(di : Ai Ai+1 ))
Ker(di : B i B i+1 ), and similarly for Im(di1 ). Thus we obtain a functor
H i : CoCh(A) A.
Definition 12.10. Let A be an abelian category.
28
HOMOLOGICAL ALGEBRA
(1) A morphism of cochain complexes f : A B of A is called a quasiisomorphism if the induced maps H i (f ) : H i (A ) H i (B ) is an isomorphism for all i Z.
(2) A cochain complex A is called acyclic if all of its cohomology objects
H i (A ) are zero.
Lemma 12.11. Let A be an abelian category.
(1) If the maps f, g : A B are homotopic, then the induced maps H i (f )
and H i (g) are equal.
(2) If f : A B is a homotopy equivalence, then f is a quasi-isomorphism.
Proof. Omitted.
H i (A )
H i+1 (A )
...
...
/ H i (B )
/ H i (C )
/ H i+1 (B )
/ H i+1 (C )
...
...
... s
Proof. Omitted. The maps come from the Snake Lemma 5.17 applied to the
diagrams
Ai /Im(di1
A )
diA
/ Ker(di+1 )
A
/ B i /Im(di1 )
B
/ C i /Im(di1 )
C
diB
/ Ker(di+1 )
B
/0
diC
/ Ker(di+1 )
C
...
/0
/ An
/ An1
/ ...
A
...
/ An+1
/ An
/ An1
/ ...
HOMOLOGICAL ALGEBRA
29
...
/ An+1
/ An
/ An1
/ ...
n A
...
/ An+1
/ An
/0
/ ...
...
/ An+1
/ Ker(dn )
/0
/ ...
A
...
/ An+1
/ An
/ An1
/ ...
...
/ An+1
/ An
/ An1
/ ...
n A
...
/0
/ Coker(dn+1 )
/ An1
/ ...
...
/0
/ An
/ An+1
/ ...
A
...
/ An1
/ An
/ An+1
/ ...
...
/ An1
/ An
/ An+1
/ ...
n A
...
/ An1
/ An
/0
/ ...
30
HOMOLOGICAL ALGEBRA
...
/ An1
/ Ker(dn )
/0
/ ...
A
...
/ An1
/ An
/ An+1
/ ...
H (n A ) =
0
if i > n
(4) The canonical truncation n A is defined by the picture
A
...
/ An1
/ An
/ An+1
/ ...
n A
...
/0
/ Coker(dn1 )
/ An+1
/ ...
H (n A ) =
H i (A ) if i n
14. Homotopy and the shift functor
It is an annoying feature that signs and indices have to be part of any discussion
of homological algebra2.
Definition 14.1. Let A be an additive category. Let A be a chain complex with
boundary maps dA,n : An An1 . For any k Z we define the k-shifted chain
complex A[k] as follows:
(1) we set A[k]n = An+k , and
(2) we set dA[k],n : A[k]n A[k]n1 equal to dA[k],n = (1)k dA,n+k .
If f : A B is a morphism of chain complexes, then we let f [k] : A[k] B[k]
be the morphism of chain complexes with f [k]n = fk+n .
Of course this means we have functors [k] : Ch(A) Ch(A) which mutually
commute (on the nose, without any intervening isomorphisms of functors), such
that A[k][l] = A[k + l] and with [0] = idCh(A) .
Definition 14.2. Let A be an abelian category. Let A be a chain complex with
boundary maps dA,n : An An1 . For any k Z we identify Hi+k (A )
Hi (A[k] ) via the identification Ai+k = A[k]i .
This identification is functorial in A . Note that since no signs are involved in
this definition we actually get a compatible system of identifications of all the homology objects Hik (A[k] ), which are further compatible with the identifications
A[k][l] = A[k + l] and with [0] = idCh(A) .
Let A be an additive category. Suppose that A and B are chain complexes,
a, b : A B are morphisms of chain complexes, and {hi : Ai Bi+1 } is a
homotopy between a and b. Recall that this means that ai bi = di+1 hi +hi1 di .
2I am sure you think that my conventions are wrong. If so and if you feel strongly about it
then drop me an email with an explanation.
HOMOLOGICAL ALGEBRA
31
Lemma 14.6. Notation and assumptions as in Lemma 14.4 above. Suppose {s0n :
Cn Bn } is a second choice of splittings. Write s0n = sn + in hn for some
unique morphisms hn : Cn An . The family of maps {hn : Cn A[1]n+1 } is a
homotopy between the associated morphisms (s), (s0 ) : C A[1] .
Proof. Omitted.
32
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
33
which occur in the long exact homology sequence associated to the short exact sequence of cochain complexes by Lemma 12.12.
Proof. Omitted.
and set k A = A . In this case Gr(A) is equivalent to the category of pairs (A, k)
consisting of an object A of A and a direct sum decomposition
M
A=
ki A
iZ
34
HOMOLOGICAL ALGEBRA
Lemma 15.2. Let A be an abelian category. The category of graded objects Gr(A)
is abelian.
Proof. Let f : A = (Ai ) B = (B i ) be a morphism of graded objects of A
given by morphisms f i : Ai B i of A. Then we have Ker(f ) = (Ker(f i )) and
Coker(f ) = (Coker(f i )) in the category Gr(A). Since we have Im = Coim in A we
see the same thing holds in Gr(A).
Remark 15.3 (Warning). There are abelian categories A having countable direct
sums but where countable direct sums are not exact. An example is the opposite
of the category of abelian sheaves on R. Namely, the category of abelian sheaves
on R has countable products, but countable
are not exact. For such a
L products
category the functor Gr(A) A, (Ai ) 7
Ai described above is not exact. It
is still true that Gr(A) is equivalent to the category of graded objects (A, k) of A,
but the kernel in the category of graded objects of a map : (A, k) (B, k) is
not equal to Ker() endowed with a direct sum decomposition, but rather it is the
direct sum of the kernels of the maps k i A k i B.
Definition 15.4. Let A be an additive category. If A = (Ai ) is a graded object,
then the kth shift A[k] is the graded object with A[k]i = Ak+i .
If A and B are graded objects of A, then we have
(15.4.1)
and an element of this group is sometimes called a map of graded objects homogeneous of degree k.
Given any set G we can define G-graded objects of A as the category whose objects
are A = (Ag )gG families of objects parametrized by elements of G. Morphisms
f : A B are defined as families of maps f g : Ag B g where g runs over the
elements of G. If G is an abelian group, then we can (unambiguously) define shift
functors [g] on the category of G-graded objects by the rule (A[g])g0 = Ag+g0 . A
particular case of this type of construction is when G = Z Z. In this case the
objects of the category are called bigraded objects of A. The (p, q) component of
a bigraded object A is usually denoted Ap,q . For (a, b) Z Z we write A[a, b] in
stead of A[(a, b)]. A morphism A A[a, b] is sometimes called a map of bidegree
(a, b).
16. Filtrations
A nice reference for this material is [Del71, Section 1]. (Note that our conventions
regarding abelian categories are different.)
Definition 16.1. Let A be an abelian category.
(1) A decreasing filtration F on an object A is a family (F n A)nZ of subobjects
of A such that
A . . . F n A F n+1 A . . . 0
(2) A filtered object of A is pair (A, F ) consisting of an object A of A and a
decreasing filtration F on A.
(3) A morphism (A, F ) (B, F ) of filtered objects is given by a morphism
: A B of A such that (F i A) F i B for all i Z.
(4) The category of filtered objects is denoted Fil(A).
HOMOLOGICAL ALGEBRA
35
(5) Given a filtered object (A, F ) and a subobject X A the induced filtration
on X is the filtration with F n X = X F n A.
(6) Given a filtered object (A, F ) and a surjection : A Y the quotient
filtration is the filtration with F n Y = (F n A).
(7) A filtration F on an object A is said to be finite if there exist n, m such
that F n A = A and F m A = 0.
T
(8) Given a filtered object (A, F ) we say F i A exists
if there exists a biggest
S
subobject of A contained in all F i A. We say F i A exists if there exists a
smallest subobject of A containing all F i A.
T
(9) The filtration on aSfiltered object (A, F ) is said to be separated if i F i A = 0
and exhaustive if F i A = A.
By abuse of notation we say that a morphism f : (A, F ) (B, F ) of filtered objects
is injective if f : A B is injective in the abelian category A. Similarly we say
f is surjective if f : A B is surjective in the category A. Being injective (resp.
surjective) is equivalent to being a monomorphism (resp. epimorphism) in Fil(A).
By Lemma 16.2 this is also equivalent to having zero kernel (resp. cokernel).
Lemma 16.2. Let A be an abelian category. The category of filtered objects Fil(A)
has the following properties:
(1)
(2)
(3)
(4)
It is an additive category.
It has a zero object.
It has kernels and cokernels, images and coimages.
In general it is not an abelian category.
Proof. It is clear that Fil(A) is additive with direct sum given by (A, F )(B, F ) =
(AB, F ) where F p (AB) = F p AF p B. The kernel of a morphism f : (A, F )
(B, F ) of filtered objects is the injection Ker(f ) A where Ker(f ) is endowed with
the induced filtration. The cokernel of a morphism f : A B of filtered objects
is the surjection B Coker(f ) where Coker(f ) is endowed with the quotient
filtration. Since all kernels and cokernels exist, so do all coimages and images. See
Example 3.12 for the last statement.
Definition 16.3. Let A be an abelian category. A morphism f : A B of filtered
objects of A is said to be strict if f (F i A) = f (A) F i B for all i Z.
This also equivalent to requiring that f 1 (F i B) = F i A + Ker(f ) for all i Z. We
characterize strict morphisms as follows.
Lemma 16.4. Let A be an abelian category. Let f : A B be a morphism of
filtered objects of A. The following are equivalent
(1) f is strict,
(2) the morphism Coim(f ) Im(f ) of Lemma 3.11 is an isomorphism.
Proof. Note that Coim(f ) Im(f ) is an isomorphism of objects of A, and that
part (2) signifies that it is an isomorphism of filtered objects. By the description
of kernels and cokernels in the proof of Lemma 16.2 we see that the filtration
on Coim(f ) is the quotient filtration coming from A Coim(f ). Similarly, the
filtration on Im(f ) is the induced filtration coming from the injection Im(f )
B. The definition of strict is exactly that the quotient filtration is the induced
filtration.
36
HOMOLOGICAL ALGEBRA
Lemma 16.5. Let A be an abelian category. Let f : A B be a strict monomorphism of filtered objects. Let g : A C be a morphism of filtered objects. Then
f g : A B C is a strict monomorphism.
Proof. Clear from the definitions.
Lemma 16.7. Let A be an abelian category. Let (A, F ), (B, F ) be filtered objects.
Let u : A B be a morphism of filtered objects. If u is injective then u is strict if
and only if the filtration on A is the induced filtration. If u is surjective then u is
strict if and only if the filtration on B is the quotient filtration.
Proof. This is immediate from the definition.
HOMOLOGICAL ALGEBRA
37
/B
g0
C
/ C qA B
in Fil(A). If f is strict, so is f 0 .
Proof. Set C qA B equal to Coker((1, 1) : A C B) in Fil(A). This cokernel
exists, by Lemma 16.2. It is a pushout, see Example 5.6. Note that F p (C A B) is
the image of F p C F p B. Hence
(f 0 )1 (F p (C A B)) = g(f 1 (F p B))) + F p C
Whence the last statement.
f0
g0
C
/B
g
/A
in Fil(A). If f is strict, so is f 0 .
Proof. This lemma is dual to Lemma 16.10.
38
HOMOLOGICAL ALGEBRA
Recall that we have defined the category Gr(A) of graded objects of A in Section
15. For (A, F ) in Fil(A) we may set
gr(A) = the graded object of A whose pth graded piece is grp (A)
and if A has countable direct sums, then we simply have
M
gr(A) =
grp (A)
This defines an additive functor
gr : Fil(A) Gr(A),
(A, F ) 7 gr(A).
f is strict,
the morphism Coim(f ) Im(f ) is an isomorphism,
gr(Coim(f )) gr(Im(f )) is an isomorphism,
the sequence gr(Ker(f )) gr(A) gr(B) is exact,
the sequence gr(A) gr(B) gr(Coker(f )) is exact, and
the sequence
0 gr(Ker(f )) gr(A) gr(B) gr(Coker(f )) 0
is exact.
Proof. The equivalence of (1) and (2) is Lemma 16.4. By Lemma 16.12 we see that
(4), (5), (6) imply (3) and that (3) implies (4), (5), (6). Hence it suffices to show that
(3) implies (2). Thus we have to show that if f : A B is an injective and surjective
map of finite filtered objects which induces and isomorphism gr(A) gr(B), then
f induces an isomorphism of filtered objects. In other words, we have to show
that f (F p A) = F p B for all p. As the filtrations are finite we may prove this by
HOMOLOGICAL ALGEBRA
39
/ F p+1 B
/ F pA
f
/ F pB
/ grp (A)
/0
grp (f )
/ grp (B)
/0
for each p Z the sequence grp (A) grp (B) grp (C) is exact,
for each p Z the sequence F p (A) F p (B) F p (C) is exact,
for each p Z the sequence A/F p (A) B/F p (B) C/F p (C) is exact,
the maps A B and B C are strict, and
A B C is exact (as a sequence in A).
Proof. Part (1) is immediate from the definitions. We will prove (3) by induction
on the length of the filtrations. If each of A, B, C has only one nonzero graded part,
then (3) holds as gr(A) = A, etc. Let n be the largest integer such that at least
one of F n A, F n B, F n C is nonzero. Set A0 = A/F n A, B 0 = B/F n B, C 0 = C/F n C
with induced filtrations. Note that gr(A) = F n A gr(A0 ) and similarly for B
and C. The induction hypothesis applies to A0 B 0 C 0 , which implies that
A/F p (A) B/F p (B) C/F p (C) is exact for p n. To conclude the same for
p = n + 1, i.e., to prove that A B C is exact we use the commutative diagram
0
/ F nA
/A
/ A0
/0
/ F nB
/B
/ B0
/0
/ F nC
/C
/ C0
/0
whose rows are short exact sequences of objects of A. The proof of (2) is dual. Of
course (5) follows from (2).
To prove (4) denote f : A B and g : B C the given morphisms. We know
that f (F p (A)) = Ker(F p (B) F p (C)) by (2) and f (A) = Ker(g) by (5). Hence
f (F p (A)) = Ker(F p (B) F p (C)) = Ker(g) F p (B) = f (A) F p (B) which proves
that f is strict. The proof that g is strict is dual to this.
40
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
41
/ Tr1 Er
Tr1 dr
/ Er
/ Tr Er
dr
Tr dr
/ Tr2 Er
/ ...
/A
with the property that the kernel of each arrow is the image of its predecessor. So Ker() = Im(f ), Ker(f ) = Im(g), and Ker(g) = Im().
(2) A morphism of exact couples t : (A, E, , f, g) (A0 , E 0 , 0 , f 0 , g 0 ) is given
by morphisms tA : A A0 and tE : E E 0 such that 0 tA = tA ,
f 0 tE = tA f , and g 0 tA = tE g.
Lemma 18.2. Let (A, E, , f, g) be an exact couple in an abelian category A. Set
(1)
(2)
(3)
(4)
(5)
(6)
d = g f : E E so that d d = 0,
E 0 = Ker(d)/Im(d),
A0 = Im(),
0 : A0 A0 induced by ,
f 0 : E 0 A0 induced by f ,
g 0 : A0 E 0 induced by g 1 .
Then we have
(1) Ker(d) = f 1 (Ker(g)) = f 1 (Im()),
(2) Im(d) = g(Im(f )) = g(Ker()),
(3) (A0 , E 0 , 0 , f 0 , g 0 ) is an exact couple.
Proof. Omitted.
Hence it is clear that given an exact couple (A, E, , f, g) we get a spectral sequence
by setting E1 = E, d1 = d, E2 = E 0 , d2 = d0 = g 0 f 0 , E3 = E 00 , d3 = d00 = g 00 f 00 ,
and so on.
Definition 18.3. Let A be an abelian category. Let (A, E, , f, g) be an exact
couple. The spectral sequence associated to the exact couple is the spectral sequence
(Er , dr )r1 with E1 = E, d1 = d, E2 = E 0 , d2 = d0 = g 0 f 0 , E3 = E 00 , d3 = d00 =
g 00 f 00 , and so on.
42
HOMOLOGICAL ALGEBRA
Tf
/ TA
/A
/ SE
Sf
/ SA
/A
_
T
Tf
TE
SE
y
T 1 SE
/ T 1 A
T 1 g
HOMOLOGICAL ALGEBRA
43
in this situation. The description of the map dr+1 is similar to that given in the
lemma. (It may be easier to use these explicit descriptions to prove one gets a
spectral sequence from such an exact couple.)
19. Spectral sequences: differential objects
Definition 19.1. Let A be an abelian category. A differential object of A is a pair
(A, d) consisting of an object A of A endowed with a selfmap d such that dd = 0. A
morphism of differential objects (A, d) (B, d) is given by a morphism : A B
such that d = d.
Lemma 19.2. Let A be an abelian category. The category of differential objects of
A is abelian.
Proof. Omitted.
/ H(A, d)
x
H(A/A, d)
where g is the canonical map and f is the map defined in the snake lemma. Thus we
get an associated spectral sequence! Since in this case we have E1 = H(A/A, d)
we see that it makes sense to define E0 = A/A and d0 = d. In other words, we
start the spectral sequence with r = 0. According to our conventions in Section 17
we define a sequence of subobjects
0 = B0 . . . Br . . . Zr . . . Z0 = E0
44
HOMOLOGICAL ALGEBRA
d1 (r A) + A
.
(r1 )1 (dA) + A
It is not hard to verify directly that (1) (3) give a spectral sequence.
Definition 19.5. Let A be an abelian category. Let (A, d) be a differential object
of A. Let : A A be an injective selfmap of A which commutes with d.
The spectral sequence associated to (A, d, ) is the spectral sequence (Er , dr )r0
described above.
Remark 19.6 (Variant). Let A be an abelian category and let S, T : A A be
shift functors, i.e., isomorphisms of categories. Assume that T S = ST as functors.
Consider pairs (A, d) consisting of an object A of A and a morphism d : A SA
such that d S 1 d = 0. The category of these objects is abelian. We define
H(A, d) = Ker(d)/Im(S 1 d) and we observe that H(SA, Sd) = SH(A, d) (canonical isomorphism). Given a short exact sequence
0 (A, d) (B, d) (C, d) 0
we obtain a long exact homology sequence
. . . S 1 H(C, d) H(A, d) H(B, d) H(C, d) SH(A, d) . . .
(note the shifts in the boundary maps). Since ST = T S the functor T defines a shift
functor on pairs by setting T (A, d) = (T A, T d). Next, let : (A, d) T 1 (A, d)
be injective with cokernel (Q, d). Then we get an exact couple as in Remark 18.5
with shift functors T S and T given by
(H(A, d), S 1 H(Q, d), , f, g)
where : H(A, d) T 1 H(A, d) is induced by , the map f : S 1 H(Q, d)
H(A, d) is the boundary map and g : H(A, d) T H(Q, d) = T S(S 1 H(Q, d))
is induced by the quotient map A T Q. Thus we get a spectral sequence as
above with E1 = S 1 H(Q, d) and differentials dr : Er T r SEr . As above we set
E0 = S 1 Q and d0 : E0 SE0 given by S 1 d : S 1 Q Q. If according to our
conventions we define Br Zr E0 , then we have for r 1 that
(1) SBr is the image of
(T r+1 . . . T 1 )1 Im(T r S 1 d)
under the natural map T 1 A Q,
(2) Zr is the image of
(S 1 T 1 d)1 Im( . . . T r1 )
under the natural map S 1 T 1 A S 1 Q.
HOMOLOGICAL ALGEBRA
45
which is given by the inclusions F A F n1 A. This is clearly an injective morphism of differential objects : (A, d) (A, d). Hence, by Definition 19.5 we get
a spectral sequence. We will call this the spectral sequence associated to the filtered
differential object (K, F, d).
Let us figure out the terms of this spectral sequence. First, note that A/A = gr(K)
endowed with its differential d = gr(d). Hence we see that
E0 = gr(K),
d0 = gr(d).
Hence the homology of the graded differential object gr(K) is the next term:
E1 = H(gr(K), gr(d)).
In addition we see that E0 is a graded object of A and that d0 is compatible with
the grading. Hence clearly E1 is a graded object as well. But it turns out that the
differential d1 does not preserve this grading; instead it shifts the degree by 1.
To work this out precisely, we define
Zrp =
F p K d1 (F p+r K) + F p+1 K
F p+1 K
and
F p K d(F pr+1 K) + F p+1 K
.
F p+1 K
This notation, although quite natural, seems to be different from the notation in
most places in the literature. Perhaps it does not matter, since the literature does
not seem to have a consistent choice of notation either. With these choices
we see
L
that Br E0 , resp. Zr E0 (as defined in Section 19) is equal to p Brp , resp.
L p
p Zr . Hence if we define
Erp = Zrp /Brp
Brp =
46
HOMOLOGICAL ALGEBRA
z + F p+1 K 7 dz + F p+r+1 K
where z F p K d1 (F p+r K).
Lemma 20.2. Let A be an abelian category. Let (K, F, d) be a filtered differential
object of A. There is a spectral sequence (Er , dr )r0 in Gr(A) associated to (K, F, d)
such that dr : Er Er [1] for all r and such that the graded pieces Erp and maps
dpr : Erp Erp+r are as given above. Furthermore, E0p = grp K, dp0 = grp (d), and
E1p = H(grp K, d).
Proof. If A has countable direct sums and if countable direct sums are exact, then
this follows from the discussion above. In general, we proceed as follows; we strongly
suggest the reader skip this proof. Consider the object A = (F p+1 K) of Gr(A), i.e.,
we put F p+1 K in degree p (the funny shift in numbering to get numbering correct
later on). We endow it with a differential d by using d on each component. Then
(A, d) is a differential object of Gr(A). Consider the map
: A A[1]
which is given in degree p by the inclusions F p+1 A F p A. This is clearly an
injective morphism of differential objects : (A, d) (A, d)[1]. Hence, we can
apply Remark 19.6 with S = id and T = [1]. The corresponding spectral sequence
(Er , dr )r0 in Gr(A) is the spectral sequence we are looking for. Let us unwind the
definitions a bit. First of all we have Er = (Erp ) is an object of Gr(A). Then, since
T r S = [r] we have dr : Er Er [r] which means that dpr : Erp Erp+r .
To see that the description of the graded pieces hold, we argue as above. Namely,
first we have E0 = Coker( : A A[1]) and by our choice of numbering above
this gives E0p = grp K. The first differential is given by dp0 = grp d : E0p E0p . Next,
the description of the boundaries Br and the cocycles Zr in Remark 19.6 translates
into a straightforward manner into the formulae for Zrp and Brp given above.
Lemma 20.3. Let A be an abelian category. Let (K, F, d) be a filtered differential
object of A. The spectral sequence (Er , dr )r0 associated to (K, F, d) has
dp1 : E1p = H(grp K) H(grp+1 K) = E1p+1
equal to the boundary map in homology associated to the short exact sequence of
differential objects
0 grp+1 K F p K/F p+2 K grp+1 K 0.
Proof. Omitted.
Definition 20.4. Let A be an abelian category. Let (K, F, d) be a filtered differential object of A. The induced filtration on H(K, d) is the filtration defined by
F p H(K, d) = Im(H(F p K, d) H(K, d)).
Lemma 20.5. Let A be an abelian category. Let (K, F, d) be a filtered differential
p
p
object of A. If Z
and B
exist (see proof ), then associated graded gr(H(K))
of the cohomology of K is a graded subquotient of the graded object E having
p
p
p
E
= Z
/B
in degree p.
HOMOLOGICAL ALGEBRA
47
p
B
Zrp =
and
=
Brp
S
=
r (F
K d1 (F p+r K) + F p+1 K)
F p+1 K
r (F
Thus
T
(F p K d1 (F p+r K) + F p+1 K)
p
E
= Sr p
.
pr+1 K) + F p+1 K)
r (F K d(F
and the top and bottom exist. On the other hand, we have
grp H(K) =
Ker(d) F p K + F p+1 K
Im(d) F p K + F p+1 K
Ker(d) F p K + F p+1 K
[
r
F p K d1 (F p+r K) + F p+1 K
and
(20.5.2)
\
r
F p K d(F pr+1 K) + F p+1 K Im(d) F p K + F p+1 K.
Definition 20.6. Let A be an abelian category. Let (K, F, d) be a filtered differential object of A. We say the spectral sequence associated to (K, F, d) converges
if gr(H(K)) = E via Lemma 20.5. In this case we also say that (Er , dr )r0 abuts
to or converges to H(K).
In the literature one finds more refined notions distinguishing between weakly
converging, abutting and converging. Namely, one can require the filtration
on H(K) to be either arbitrary, or exhaustive and separated, or exhaustive
and complete in addition to the condition that gr(H(K)) = E . We try to
avoid introducing this notation by simply adding the relevant information in the
statements of the results.
Lemma 20.7. Let A be an abelian category. Let (K, F, d) be a filtered differential
object of A. The associated spectral sequence converges if and only if for every
p Z we have equality in equations (20.5.2) and (20.5.1).
Proof. Immediate from the discussions above.
48
HOMOLOGICAL ALGEBRA
L n
us denote d the differential of K. Forgetting the grading we can think of
K
as a filtered differential object of A. Hence according to Section 20 we obtain
a spectral sequence (Er , dr )r0 . In this section we work out the terms of this
spectral sequence, and we endow the terms of this spectral sequence with additional
structure coming from the grading of K.
First we point out that E0p = grp K is a complex and hence is graded. Thus E0 is
bigraded in a natural way. It is customary to use the bigrading
M
E0 =
E0p,q , E0p,q = grp K p+q
p,q
The idea is that p + q should be thought of as the total degree of the (co)homology
classes. Also, p is called the filtration degree, and q is called the complementary
degree. The differential d0 is compatible with this bigrading in the following way
M p,q
p,q
p,q+1
d0 , dp,q
.
d0 =
0 : E0 E0
Namely, dp0 is just the differential on the complex grp K (which occurs as grp E0
just shifted a bit).
To go further we identify the objects Brp and Zrp introduced in Section 20 as graded
objects and we work out the corresponding decompositions of the differentials. We
do this in a completely straightforward manner, but again we warn the reader that
our notation is not the same as notation found elsewhere. We define
Zrp,q =
and
F p K p+q d(F pr+1 K p+q1 ) + F p+1 K p+q
.
F p+1 K p+q
and ofL
course Erp,q = L
Zrp,q /Brp,q . With these
L definitions it is completely clear that
p
p
p,q
Zr = q Zr , Br = q Brp,q , and Erp = q Erp,q . Moreover,
Brp,q =
HOMOLOGICAL ALGEBRA
49
50
HOMOLOGICAL ALGEBRA
p,q
the cohomology of K is a graded subquotient of the graded object p+q=n E
.
Proof. Let q = n p. As in the proof of Lemma 20.5 we see that
T
(F p K n d1 (F p+r K n+1 ) + F p+1 K n )
p,q
E
= Sr p n
.
pr+1 K n1 ) + F p+1 K n )
r (F K d(F
On the other hand, we have
grp H n (K) =
(21.6.1)
Ker(d) F p K n + F p+1 K n
Im(d) F p K n + F p+1 K n
[
r
and
(21.6.3)
\
r
F p K n d(F pr+1 K n1 ) + F p+1 K n Im(d) F p K n + F p+1 K n .
d
dp,q
1
2 = d2
1
for all p, q Z.
HOMOLOGICAL ALGEBRA
51
This is just the cochain version of the definition. It says that each Ap, is a cochain
complex and that each dp,
is a morphism of complexes Ap, Ap+1, such that
1
p,
p+1,
d1 = 0 as morphisms of complexes. In other words a double complex can
d1
be seen as a complex of complexes. So in the diagram
...
. .O .
...
/ Ap,q+1
O
. .O .
...
/ Ap+1,q+1
O
/ ...
dp,q+1
1
dp,q
2
dp+1,q
2
/ Ap,q
O
...
dp,q
1
/ Ap+1,q
O
/ ...
...
...
...
...
any square commutes. Warning: In the literature one encounters a different definition where a bicomplex or a double complex has the property that the squares
in the diagram anti-commute.
Example 22.2. Let A, B, C be abelian categories. Suppose that
: A B C,
(X, Y ) 7 X Y
is a functor which is bilinear on morphisms, see Categories, Definition 2.20 for the
definition of A B. Given a complexes X of A and Y of B we obtain a double
complex
K , = X Y
in C. Here the first differential K p,q K p+1,q is the morphism X p Y q
X p+1 Y q induced by the morphism X p X p+1 and the identity on Y q . Similarly
for the second differential.
Let A, be a double complex. It is customary to denote HIp (A, ) the complex
p1,q
with terms Ker(dp,q
) (varying q) and differential induced by d2 . Then
1 )/Im(d1
p
q
,
HII (HI (A )) denotes its cohomology in degree q. It is also customary to denote
p,q1
q
(A, ) the complex with terms Ker(dp,q
) (varying p) and differential
HII
2 )/Im(d2
p
q
,
induced by d1 . Then HI (HII (A )) denotes its cohomology in degree q. It will turn
out that these cohomology groups show up as the terms in the spectral sequence
for a filtration on the associated to total complex.
Definition 22.3. Let A be an additive category. Let A, be a double complex.
The associated simple complex sA , also sometimes called the associated total complex is given by
M
sAn =
Ap,q
n=p+q
X
n=p+q
p p,q
(dp,q
1 + (1) d2 )
52
HOMOLOGICAL ALGEBRA
There are two natural filtrations on the simple complex sA associated to the double
complex A, . Namely, we define
M
M
p
FIp (sAn ) =
Ai,j and FII
(sAn ) =
Ai,j .
i+j=n, ip
i+j=n, jp
It is immediately verified that (sA , FI ) and (sA , FII ) are filtered complexes. By
Section 21 we obtain two spectral sequences. It is customary to denote (0 Er , 0 dr )r0
the spectral sequence associated to the filtration FI and to denote (00 Er , 00 dr )r0
the spectral sequence associated to the filtration FII . Here is a description of these
spectral sequences.
Lemma 22.4. Let A be an abelian category. Let K , be a double complex. The
spectral sequences associated to K , have the following terms:
p,q
p p,q
K p,q+1 ,
(1) 0 E0p,q = K p,q with 0 dp,q
0 = (1) d2 : K
p,q
p,q
q,p
(2) 00 E0 = K q,p with 00 d0 = d1 : K q,p K q+1,p ,
q p,
(3) 0 E1p,q = H q (K p, ) with 0 dp,q
1 = H (d1 ),
p,q
p,q
(4) 00 E1 = H q (K ,p ) with 00 d1 = (1)q H q (d,p
2 ),
q
(5) 0 E2p,q = HIp (HII
(K , )),
p
(6) 00 E2p,q = HII
(HIq (K , )).
Proof. Omitted.
These spectral sequences define two filtrations on H n (sK ). We will denote these
FI and FII .
Definition 22.5. Let A be an abelian category. Let K , be a double complex.
We say the spectral sequence (0 Er , 0 dr )r0 converges if Definition 21.7 applies. In
other words, for all n
p,q
grFI (H n (sK )) = p+q=n 0 E
via the canonical comparison of Lemma 21.6. Similarly we say the spectral sequence
(00 Er , 00 dr )r0 converges if Definition 21.7 applies. In other words for all n
p,q
grFII (H n (sK )) = p+q=n 00 E
HOMOLOGICAL ALGEBRA
53
(4) The complex Ap, is exact in all degrees q 6= 0 and the morphism K p Ap,0
induces an isomorphism K p Ker(dp,0
2 ).
Then induces a quasi-isomorphism
K sA
of complexes. Moreover, there is a variant of this lemma involving the second
variable q instead of p.
Proof. The map is simply the map given by the morphisms K n An,0 sAn ,
which are easily seen to define a morphism of complexes. Consider the spectral
sequence (0 Er , 0 dr )r0 associated to the double complex A, . By Lemma 22.6 this
spectral sequence converges and the induced filtration on H n (sA ) is finite for each
n. By Lemma 22.4 and assumption (4) we have 0 E1p,q = 0 unless q = 0 and 0 E1p,0 =
K p with differential 0 dp,0
identified with dpK . Hence 0 E2p,0 = H p (K ) and zero
1
p,q
otherwise. This clearly implies dp,q
2 = d3 = . . . = 0 for degree reasons. Hence we
n
0 p,q
in C. By induction we see that each E is in C. Hence each H (sK ) has a finite
filtration whose subquotients are in C. Using that C is closed under extensions we
conclude that H n (sK ) is in C as claimed.
The same result holds for the second spectral sequence associated to K , . Similarly,
if (K , F ) is a filtered complex to which Lemma 21.9 applies and for some r 0
all the objects Erp,q belong to C, then each H n (K ) is an object of C.
Remark 22.9. Let A be an additive category. Let A,, be a triple complex. The
associated total complex is the complex with terms
M
Totn (A,, ) =
Ap,q,r
p+q+r=n
and differential
dnTot(A,, ) =
X
p+q+r=n
+ (1)p+q dp,q,r
dp,q,r
+ (1)p dp,q,r
1
2
3
With this definition a simple calculation shows that the associated total complex
is equal to
Tot(A,, ) = Tot(Tot12 (A,, )) = Tot(Tot23 (A,, ))
In other words, we can either first combine the first two of the variables and then
combine sum of those with the last, or we can first combine the last two variables
and then combine the first with the sum of the last two.
Lemma 22.10. Let M be a complex of abelian groups. Let
. . . A2 A1 A0 M 0
54
HOMOLOGICAL ALGEBRA
be an exact complex of complexes of abelian groups such that for all p Z the
complexes
. . . Ker(dpA ) Ker(dpA ) Ker(dpA ) Ker(dpM ) 0
2
Aqp
M induced by A0 M is a quasi-isomorphism.
p,q
Proof. Write TP
= Tot(A, ). Let x Ker(d0T ) represent a cohomology class
i
. Write x =
i=n,...,0 xi with xi Ai . Assume n > 0. Then xn is in the
n
kernel of dAn and maps to zero in the cohomology of An1 (because it maps to an
element which is the boundary of xn1 up to sign). The condition on exactness
of kernels of differentials implies that the cohomology class of xn is in the image
of H n (An+1 ) H n (An ) (details omitted). Thus we can modify x by a boundary
and reach the situation where xn is a boundary. Modifying x once more we see
that we may assume xn = 0. By induction we see that every cohomology class
is represented by a cocycle x = x0 . Finally, the condition on exactness of kernels
tells us two such cocycles x0 and x00 are cohomologous if and only if their image in
H 0 (M ) are the same.
Aqp
p+q=n
Aqp
As differential we use
d((xp,q )) = (fp (xp1,q ) + (1)p dAp (xp,q1 ))
Let x Ker(d0T ) represent a cohomology class H 0 (T ). Write x = (xi ) with
1
xi Ai
A01 ). Hence we see that
i . Note that x0 maps to zero in Coker(A1
x0 = m0 + d(y) for some m0 M 0 . Then d(m0 ) = 0 because d(x0 ) = 0 as x is
a cocycle. Thus, replacing by something in the image of H 0 (M ) H 0 (T ) we
0
may assume that x0 is in the image of d : A1
0 A0 .
0
Assume x0 Im(A1
0 A0 ). We claim that in this case = 0. To prove this we
find, by induction on n elements y1 , . . . , yn with yi Ai1
such that x0 = d(y0 )
i
and xj = fj1 (yj1 ) + (1)j d(yj ). This is clear for n = 0. Proof of induction step
is omitted. Taking y = (yi ) we find that d(y) = .
HOMOLOGICAL ALGEBRA
55
23. Injectives
Definition 23.1. Let A be an abelian category. An object J Ob(A) is called
injective if for every injection A , B and every morphism A J there exists a
morphism B J making the following diagram commute
/B
A
J
Here is the obligatory characterization of injective objects.
Lemma 23.2. Let A be an abelian category. Let I be an object of A. The following
are equivalent:
(1) The object I is injective.
(2) The functor B 7 HomA (B, I) is exact.
(3) Any short exact sequence
0IAB0
in A is split.
(4) We have ExtA (B, I) = 0 for all B Ob(A).
Proof. Omitted.
56
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
57
but v does not transform injective maps into injective maps in general (i.e., v is
not left exact). Moreover, it is not the case that u transforms injective B-modules
into injective A-modules. For example, if A = Z and B = Z/pZ, then the injective
B-module Z/pZ is not an injective Z-module. In fact, the lemma applies to this
example if and only if the ring map A B is flat.
Lemma 25.3. Let A and B be abelian categories. Let u : A B and v : B A
be additive functors. Assume
(1)
(2)
(3)
(4)
u is right adjoint to v,
v transforms injective maps into injective maps,
A has enough injectives, and
vB = 0 implies B = 0 for any B Ob(B).
u is right adjoint to v,
v transforms injective maps into injective maps,
A has enough injectives,
vB = 0 implies B = 0 for any B Ob(B), and
A has functorial injective hulls.
7 u() iP
58
HOMOLOGICAL ALGEBRA
7 v()
which is characterized by the rule u(v()) iP2 = iP1 in HomB (P2 , u(v(P1 ))).
Note that 7 v() is compatible with composition; this can be seen directly from
the characterization. Hence P 7 v(P ) is a functor from the full subcategory of B
whose objects are the elements of P.
Given an arbitrary object B of B choose an exact sequence
P2 P 1 B 0
which is possible by assumption (1). Define v(B) to be the object of A fitting into
the exact sequence
v(P2 ) v(P1 ) v(B) 0
Then
HomA (v(B), A) = Ker(HomA (v(P1 ), A) HomA (v(P2 ), A))
= Ker(HomB (P1 , u(A)) HomB (P2 , u(A)))
= HomB (B, u(A))
Hence we see that we may take P = Ob(B), i.e., we see that v is everywhere
defined.
26. Essentially constant systems
In this section we discuss essentially constant systems with values in additive categories.
Lemma 26.1. Let I be a category, let A be a pre-additive Karoubian category, and
let M : I A be a diagram.
(1) Assume I is filtered. The following are equivalent
(a) M is essentially constant,
(b) X = colim M exists and there exists a cofinal filtered subcategory I 0
I and for i0 Ob(I 0 ) a direct sum decomposition Mi0 = Xi0 Zi0 such
that Xi0 maps isomorphically to X and Zi0 to zero in Mi00 for some
i0 i00 in I 0 .
(2) Assume I is cofiltered. The following are equivalent
(a) M is essentially constant,
(b) X = lim M exists and there exists an initial cofiltered subcategory I 0
I and for i0 Ob(I 0 ) a direct sum decomposition Mi0 = Xi0 Zi0 such
that X maps isomorphically to Xi0 and Mi00 Zi0 is zero for some
i00 i0 in I 0 .
Proof. Assume (1)(a), i.e., I is filtered and M is essentially constant. Let X =
colim Mi . Choose i and X Mi as in Categories, Definition 22.1. Let I 0 be the full
subcategory consisting of objects which are the target of a morphism with source
i. Suppose i0 Ob(I 0 ) and choose a morphism i i0 . Then X Mi Mi0
composed with Mi0 X is the identity on X. As A is Karoubian, we find a direct
summand decomposition Mi0 = Xi0 Zi0 , where Zi0 = Ker(Mi0 X) and Xi0
maps isomorphically to X. Pick i k and i0 k such that Mi0 X Mi Mk
equals Mi0 Mk as in Categories, Definition 22.1. Then we see that Mi0 Mk
annihilates Zi0 . Thus (1)(b) holds.
HOMOLOGICAL ALGEBRA
59
Assume (1)(b), i.e., I is filtered and we have I 0 I and for i0 Ob(I 0 ) a direct sum
decomposition Mi0 = Xi0 Zi0 as stated in the lemma. To see that M is essentially
constant we can replace I by I 0 , see Categories, Lemmas 22.8 and 17.2. Pick any
i Ob(I) and denote X Mi the inverse of the isomorphism Xi X followed
by the inclusion map Xi Mi . If j is a second object, then choose j k such
that Zj Mk is zero. Since I is filtered we may also assume there is a morphism
i k (after possibly increasing k). Then Mj X Mi Mk and Mj Mk
both annihilate Zj . Thus after postcomposing by a morphism Mk Ml which
annihilates the summand Zk , we find that Mj X Mi Ml and Mj Ml
are equal, i.e., M is essentially constant.
The proof of (2) is dual.
M1 M2 M3 . . .
Moreover, we often omit the transition maps i from the notation and we simply
say let (Mi ) be an inverse system.
60
HOMOLOGICAL ALGEBRA
The collection of all inverse systems with values in C forms a category with the
obvious notion of morphism.
Lemma 27.1. Let C be a category.
(1) If C is an additive category, then the category of inverse systems with values
in C is an additive cateogry.
(2) If C is an abelian category, then the category of inverse systems with values
in C is an abelian cateogry. A sequence (Ki ) (Li ) (Mi ) of inverse
systems is exact if and only if each Ki Li Ni is exact.
Proof. Omitted.
The limit (see Categories, Section 21) of such an inverse system is denoted lim Mi ,
or limi Mi . If C is the category of abelian groups (or sets), then the limit always
exists and in fact can be described as follows
Y
limi Mi = {(xi )
Mi | i (xi ) = xi1 , i = 2, 3, . . .}
see Categories, Section 15. However, given a short exact sequence
0 (Ai ) (Bi ) (Ci ) 0
of inverse systems of abelian groups it is not always the case that the associated
system of limits is exact. In order to discuss this further we introduce the following
notion.
Definition 27.2. Let C be an abelian category. We say the inverse system (Ai )
satisfies the Mittag-Leffler condition, or for short is ML, if for every i there exists
a c = c(i) i such that
Im(Ak Ai ) = Im(Ac Ai )
for all k c.
It turns out that the Mittag-Leffler condition is good enough to ensure that the
lim-functor is exact, provided one works within the abelian category of abelian
groups, or abelian sheaves, etc. It is shown in a paper by A. Neeman (see [Nee02])
that this condition is not strong enough in a general abelian category (where limits
of inverse systems exist).
Lemma 27.3. Let
0 (Ai ) (Bi ) (Ci ) 0
be a short exact sequence of inverse systems of abelian groups.
(1) In any case the sequence
0 limi Ai limi Bi limi Ci
is exact.
(2) If (Bi ) is ML, then also (Ci ) is ML.
(3) If (Ai ) is ML, then
0 limi Ai limi Bi limi Ci 0
is exact.
Proof. Nice exercise. See Algebra, Lemma 84.1 for part (3).
HOMOLOGICAL ALGEBRA
61
/ Ac0
/ Bc0
/ C Zc 0
/0
/ Ac
/ Bc
/ C Zc
/0
/ Ai
/ Bi
/ C Zi
/0
62
HOMOLOGICAL ALGEBRA
HOMOLOGICAL ALGEBRA
63
29. Other chapters
Preliminaries
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
Introduction
Conventions
Set Theory
Categories
Topology
Sheaves on Spaces
Sites and Sheaves
Stacks
Fields
Commutative Algebra
Brauer Groups
Homological Algebra
Derived Categories
Simplicial Methods
More on Algebra
Smoothing Ring Maps
Sheaves of Modules
Modules on Sites
Injectives
Cohomology of Sheaves
Cohomology on Sites
Differential Graded Algebra
Divided Power Algebra
Hypercoverings
Schemes
(25) Schemes
(26) Constructions of Schemes
(27) Properties of Schemes
(28)
(29)
(30)
(31)
(32)
(33)
(34)
(35)
(36)
(37)
(38)
(39)
(40)
Morphisms of Schemes
Cohomology of Schemes
Divisors
Limits of Schemes
Varieties
Topologies on Schemes
Descent
Derived Categories of Schemes
More on Morphisms
More on Flatness
Groupoid Schemes
More on Groupoid Schemes
Etale
Morphisms of Schemes
Chow Homology
Adequate Modules
Dualizing Complexes
Etale
Cohomology
Crystalline Cohomology
Pro-etale Cohomology
Algebraic Spaces
(47)
(48)
(49)
(50)
(51)
(52)
(53)
Algebraic Spaces
Properties of Algebraic Spaces
Morphisms of Algebraic Spaces
Decent Algebraic Spaces
Cohomology of Algebraic Spaces
Limits of Algebraic Spaces
Divisors on Algebraic Spaces
64
HOMOLOGICAL ALGEBRA
(54)
(55)
(56)
(57)
(58)
(59)
(60)
(61)
(62)
Topics in Geometry
(63)
(64)
(65)
(66)
(67)
Quotients of Groupoids
Simplicial Spaces
Formal Algebraic Spaces
Restricted Power Series
Resolution of Surfaces
Deformation Theory
(68) Formal Deformation Theory
(69) Deformation Theory
(70) The Cotangent Complex
Algebraic Stacks
References
Pierre Deligne, Th
eorie de Hodge. II, Inst. Hautes Etudes
Sci. Publ. Math. (1971), no. 40,
557.
[Eis95] David Eisenbud, Commutative algebra, Graduate Texts in Mathematics, vol. 150,
Springer-Verlag, 1995.
[Lan02] Serge Lang, Algebra, third ed., Graduate Texts in Mathematics, vol. 211, SpringerVerlag, New York, 2002.
[McC01] John McCleary, A users guide to spectral sequences, second ed., Cambridge Studies in
Advanced Mathematics, vol. 58, Cambridge University Press, Cambridge, 2001.
[ML63] Saunders Mac Lane, Homology, Die Grundlehren der mathematischen Wissenschaften,
Bd. 114, Academic Press Inc., Publishers, New York, 1963.
[Nee02] Amnon Neeman, A counterexample to a 1961 theorem in homological algebra, Invent.
Math. 148 (2002), no. 2, 397420, With an appendix by P. Deligne.
[Ser53] Jean-Pierre Serre, Groupes dhomotopie et classes de groupes ab
eliens, Ann. of Math.
(2) 58 (1953), 258294.
[Del71]